1. News
  2. Everyday tech
  3. OpenAI · Elections
OpenAI · Elections2 min read

ChatGPT now turns fake election photos into cartoons

ChatGPT now turns requests for fake election photos into cartoons, which makes them easier to spot before US midterm voters share them.

By the Dashline deskSource: Tom's Guide
ChatGPT now turns fake election photos into cartoons
Photo: Unknown authorUnknown author, CC0

Ahead of next week's US midterm elections, ChatGPT will draw cartoons instead of realistic photos when someone asks for images tied to election interference. A Tom's Guide writer tested it. He asked for a photorealistic picture of someone stuffing mail ballots into a drop box and got an illustration. Asking again did not change that.

What changed

OpenAI, the company behind ChatGPT, described the change in a May blog post titled "Election information and safeguards in 2026." It applies image stylization to requests for images that show election interference or that may discourage people from voting. The results are stylized, or cartoonified, illustrations. OpenAI says the goal is to reduce the risk that an imagined scene is mistaken for documentary evidence.

OpenAI also says its rules against election interference and voter discouragement still apply when the image is a cartoon. The approach leaves room for illustrations and social commentary.

What the test showed

The reviewer said he asked for a photorealistic style in his first request, but the image still came back as an illustration. He then asked again for a photorealistic version. ChatGPT only polished the original cartoon and stayed in cartoon form.

Why it matters

A realistic fake photo of ballots being dumped, or of a candidate doing something they never did, can fool people quickly. The reviewer wrote that he has seen many older people in comment sections fall for obvious political deepfakes. A cartoon look is a built-in clue that the picture is not real evidence.

What other AI companies are doing

OpenAI is one of several companies that signed the 2024 "Tech Accord to Combat Deceptive Use of AI in 2024 Elections." Others include Anthropic, Meta, Google and Microsoft. For 2026, Google and Anthropic are focused on pointing voters to reliable sources.

  • Google says it is adding election tools in Search, AI Mode, AI Overviews and the Gemini app.
  • Those tools cover polling locations and registration deadlines from state and local governments and Democracy Works.
  • Google also plans real-time voting results from The Associated Press.
  • Anthropic's Claude shows a banner linking to TurboVote, a nonpartisan resource from Democracy Works.
  • Anthropic says Claude will start using official Democracy Works voting information in its answers late this month.

What is not known

The article reports one reviewer's test of one kind of request. It does not explain how ChatGPT decides which requests count as election interference. It also does not say how well the cartoon rule holds up across many prompts, or whether image tools from other companies work the same way.

The takeaway

A realistic election image is not proof of anything, and ChatGPT now draws cartoons for election-interference requests, but check official sources for voting details.

More articles

You should really read these.

Next step

Let’s talk